Passive Sampling to Measure Baseline Dissolved Persistent Organic

Oct 12, 2012 - In September 2010, PE and SPME samplers, prepared as described above, were codeployed at multiple stations along the 45 and 60 m ...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/est

Passive Sampling to Measure Baseline Dissolved Persistent Organic Pollutant Concentrations in the Water Column of the Palos Verdes Shelf Superfund Site Loretta A. Fernandez,*,† Wenjian Lao,‡ Keith A. Maruya,‡ Carmen White,§ and Robert M. Burgess† †

U.S. Environmental Protection Agency, Atlantic Ecology Division, Narragansett, Rhode Island 02882, United States Southern California Coastal Water Research Project (SCCWRP), Costa Mesa, California 92626, United States § U.S. Environmental Protection Agency, Region 9, San Francisco, California 94109, United States ‡

S Supporting Information *

ABSTRACT: Passive sampling was used to deduce water concentrations of persistent organic pollutants (POPs) in the vicinity of a marine Superfund site on the Palos Verdes Shelf, California, USA. Precalibrated solid phase microextraction (SPME) fibers and polyethylene (PE) strips that were preloaded with performance reference compounds (PRCs) were codeployed for 32 d along an 11-station gradient at bottom, surface, and midwater depths. Retrieved samplers were analyzed for DDT congeners and their breakdown products (DDE, DDD, DDMU, and DDNU) and 43 PCB congeners using GC-EI- and NCI-MS. PRCs were used to calculate compound-specific fractional equilibration achieved in situ for the PE samplers, using both an exponential approach to equilibrium (EAE) and numerical integration of Fickian diffusion (NI) models. The highest observed concentrations were for p,p′-DDE, with 2200 and 990 pg/L deduced from PE and SPME, respectively. The difference in these estimates could be largely attributed to uncertainty in equilibrium partition coefficients, unaccounted for disequilibrium between samplers and water, or different time scales over which the samplers average. The concordance between PE and SPME estimated concentrations for DDE was high (R2 = 0.95). PCBs were only detected in PE samplers, due to their much larger size. Near-bottom waters adjacent to and down current from sediments with the highest bulk concentrations exhibited aqueous concentrations of DDTs and PCBs that exceeded Ambient Water Quality Criteria (AWQC) for human and aquatic health, indicating the need for future monitoring to determine the effectiveness of remedial activities taken to reduce adverse effects of contaminated surface sediments.

■

INTRODUCTION

Protection Agency has explored remedial alternatives for the site.8 Capping of the most contaminated sediments near the outfall has been selected as the preferred remedial alternative.8 Monitoring of POP concentrations in the water column before (i.e., baseline conditions), during, and after the capping is necessary to evaluate the effectiveness of the remediation as well as any adverse side effects (e.g., resuspension-caused contaminant release). Measurement of dissolved concentrations ranging from femtograms per liter to tens of nanograms per liter for individual compounds, however, is technically challenging and time-consuming using traditional techniques, requiring pumping and filtering of large volumes of water (hundreds to thousands of liters) in order to collect analytically detectable contaminant masses on sorptive material (e.g., octadecyl resin or polyurethane foam).12,14,15

Sediments of the Palos Verdes Shelf (PVS) off the coast of California are heavily contaminated with persistent organic pollutants (POPs), specifically, organochlorine pesticides (including dichlorodiphenyltrichloroethane (DDT) and its breakdown products) and polychlorinated biphenyls (PCBs).1−6 Swartz et al.7 found the presence of these contaminants likely contributed to elevated whole sediment acute toxicity to marine amphipods exposed to sediments from PVS. These contaminants are primarily the legacy of industrial wastes released to the shelf through the outfall of the Joint Water Pollution Control Plant (JWPCP), a wastewater treatment facility operated by the Los Angeles County Sanitation Districts (LACSD).8 The prevailing coastal currents and wave activity carry remnants of the discharge in a northwesterly direction from the outfall, located off White Point (Figure 1).9 As for many contaminated sites, sediments on the shelf continue be a source of chemicals to the water column, and biota living within and above it, long after the most polluted industrial releases have ended.10−13 To protect human and ecological health, the U.S. Environmental © 2012 American Chemical Society

Received: Revised: Accepted: Published: 11937

May 31, October October October

2012 4, 2012 12, 2012 12, 2012

dx.doi.org/10.1021/es302139y | Environ. Sci. Technol. 2012, 46, 11937−11947

Environmental Science & Technology

Article

Because equilibrium between PE samplers and water cannot always be assumed for very hydrophobic compounds, such as some DDTs (e.g., DDD, DDE, and DDT) and highly chlorinated PCBs, following short deployments (days to weeks), performance reference compounds (PRCs) are used to correct for nonattainment of equilibrium conditions.17,19 Similarly to SPME, C W may be calculated from the concentration of analyte taken up by the PE strips, while accounting for disequilibrium using PRCs that are similar to the analyte in terms of diffusivities and partitioning coefficient:

In situ passive sampling methods, including solid-phase microextraction (SPME) and low-density polyethylene (PE) strips reduce the labor involved in sampling and postcollection processing, while allowing for dissolved POP measurement at very low concentrations. Further, passive sampling avoids many of the artifacts associated with traditional methods that result in over- and under-estimating dissolved concentrations. In general, passive samplers allow dissolved concentrations to be deduced from concentrations in equilibrated polymeric phases using polymer−water partition coefficients (e.g., KPEW and Kf).16,17 Zeng et al.16 applied SPMEs, polydimethylsiloxane (PDMS) coated glass fibers, to measure DDT breakdown products (p,p′DDE, o,p′-DDE, and p,p′-DDD) dissolved in the water column of the PVS. Concentrations in the range of tens of picograms per liter to nanograms per liter matched those measured using the traditional pumping, filtering, and solid phase extraction of 999 L of seawater. SPME were believed to be equilibrated with the PVS water DDTs after 18 d deployments, as no statistically significant differences in concentration were seen in SPME exposed for 18 and 30 d. Similarly, Adams18 and Adams et al.17 used PE strips, exposed for 15 d and assuming sampler/water equilibration, to sample PCB congener #52 in Boston Harbor and lower Hudson River estuary in the range of pictograms per liter. To monitor the effectiveness of remedial efforts on the PVS, water concentrations for individual PCB congeners will have to be detectable to the range of tens of femtograms per liter. In this work, both SPME and PE passive samplers are used simultaneously to measure selected DDT and PCB concentrations at eleven stations along the PVS and at a background station with historically lower detectable POP concentrations. By using large strips of PE, it is expected that even highly chlorinated PCBs, which have previously been measured in the sediments, but not detectable in the water column due to extremely low solubility, could be measured (expected range of tens to hundreds of femtograms per liter). These data will allow benchmark, time-averaged concentrations to be set for the PVS, to which water column concentrations, measured using similar methods during and after remedial activity at the site, may be compared. Co-deployment of the two types of samplers will allow for a direct comparison of estimated concentrations providing confidence in the techniques where results are consistent, and highlighting necessary adjustments to their methods of use where results are inconsistent. Finally, an alternative method for calibrating the fractional equilibration of membrane samplers is presented and compared with a frequently used, existing method. This alternative method may be used with any sampler material and thickness and may be useful in predicting fractional equilibrations before deployment, helping to improve the design of sampling campaigns.

■

(2)

C ∞PE = CPE , t /feq

(3)

and feq = (C 0 PRC − CPRC , t )/C 0 PRC

(4)

C∞PE

where is the equilibrium analyte concentration in the PE in ng/kgPE, KPEW is the compound-specific polyethylene−water partition coefficient (LW/kgPE), CPE,t is the concentration in the PE after deployment, C0PRC is the initial concentration of PRC in the sampler, CPRC,t is the concentration of PRC in the sampler after deployment, and feq is the fractional equilibration of the sampler to the water. Challenges to using PRCs are encountered when wishing to interpolate between, or extrapolate from, PRCs to compounds that differ in terms of diffusivity and partitioning behavior. Existing methods are based on an exponential approach to equilibrium (EAE) models that were originally developed for semipermeable membrane devices (SPMDs).20,21 Constant mass transfer coefficients are assumed and uptake/release by polymeric passive samplers is often described as being either polymer-side controlled, or water-side boundary layer controlled, depending on the diffusion rates across each layer. While semiempirical models for specific samplers have been developed to predict the contributions to mass-transfer resistance of the polymer and water-side boundary layers,22,23 a more generic method, which could be used with any sampler material and thickness, may simplify the application of polymer films as passive samplers. An alternative to using EAE models may be found in the numerical integration of explicit, finite-difference modeling of Fickian, mass diffusion across both polymer and water layers, while accounting for the no flux boundary at the center of the polymer film (NI method). Using a stagnant film model to describe the water-side boundary layer (BL), one-dimensional Fickian diffusion in a system consisting of a PE sheet, BL, and a well mixed infinite bath is described in the Supporting Information (following Crank24). While polymer membrane thicknesses are known, BL thicknesses are unknown for any given deployment, and can change with flow rates and turbulence.20,21 However, data from multiple PRCs may be used to find the best fit modeled BL. The same model, with a best-fit BL, may then be used to find feq for compounds which do not have matching PRCs (complete description of model in Supporting Information).24 For the purposes of this work, BL dependence on a compound’s diffusivity in water is neglected.

THEORY

Dissolved concentrations, CW, may be deduced using data from both types of passive samplers. First, SPME-water partition coefficients, Kf (LW/LPDMS), may be used to calculate CW (ng/ L) from the mass of analyte sorbed to the fiber, Nf (ng), assuming the fibers and water are fully equilibrated: CW = Nf /(K f Vf )

CW = C ∞PE /KPEW

■

MATERIALS AND METHODS Passive Sampler Preparation. PE strips (10 cm ×100 cm) were prepared from low-density polyethylene sheets (25 μm, ACE Hardware Corp., Oak Brook, IL, USA) and solvent

(1)

where Vf (LPDMS) is the volume of the sorptive coating. 11938

dx.doi.org/10.1021/es302139y | Environ. Sci. Technol. 2012, 46, 11937−11947

Environmental Science & Technology

Article

Figure 1. Black triangles indicate sampling stations off Palos Verdes, California (USA). PE and SPME were codeployed at stations BA1C, BA3C, BA4C, BA5C, BA7C, BA7DC, BA8C, BA8DC, BA9C, and T11 (not shown). Only PE samplers were deployed at stations BA5DC and BA9DC.

cleaned as described previously.25 Each sampler was then soaked in an aqueous PRC solution (13C-labeled p,p′-DDT, p,p′-DDE, p,p′-DDD, PCB#28, PCB#52, PCB#118, and PCB#128 (Cambridge Isotope, Andover, MA, USA)) in a 1L amber glass jar for at least 20 weeks before deployment. To streamline analysis, target PRC concentrations in PE were approximately ten times the expected concentrations of target chemicals so that all concentrations would fall within the same range, assuming up to 90% loss of PRC during deployment. Finally, samplers were threaded onto solvent-rinsed aluminum wires, wrapped in aluminum foil, and stored at −20 °C or on ice until deployed. SPME (100-μm PDMS-coated silica fibers, Supelco, Bellefonte, PA, USA) were prepared and handled as described previously.16 Briefly, newly purchased SPME fibers were preconditioned at 250 °C for 0.5 h prior to assembly into individual perforated copper casings for protection during deployment. Each fiber/casing assembly was kept in a sealed glass vial in a freezer at −20 °C and shipped on ice until deployment. Field Deployment and Retrieval. In September 2010, PE and SPME samplers, prepared as described above, were codeployed at multiple stations along the 45 and 60 m isobaths of the PVS, from just up current of the JWPCP outfall, to approximately 10 km down current (Figure 1), and at a background station (T11) approximately 24 km SE of the outfall (station labels reflect those used in previous monitoring

of PVS sediment and water). Samplers were codeployed at three depths per station: 5 m below the surface (“surface”); 5 m above the bottom (“near bottom”), and 30−35 m below the surface (“mid-depth”). PE samplers were deployed in triplicate at each depth for all 12 stations shown in Figure 1. SPME samplers were deployed in quadruplicate at each depth for 10 of the 12 stations (excluding BA5DC and BA9DC). Both types of samplers were retrieved after 32 days. Because the aluminum wire on which PE were threaded broke for the shallower samplers, only PEs from the near bottom depth were recovered from eight stations, with the addition of the mid-depth samplers at stations BA4C and BA5DC. Copper or stainless steel wire would have been more appropriate for this application. All SPME samplers were recovered. Water temperature and salinity were measured using conductivity, temperature, and depth meter (CTD) casts at time of deployment. Dissolved organic carbon (DOC) was determined on discrete water samples collected at depth using a Niskin bottle. Analyses for DOC were performed on a Shimadzu TOC-VCSH analyzer (Kyoto, Japan). All samplers were returned to the laboratory on ice and remained frozen until analyzed. Sampler Analysis. PE strips were wiped with laboratory tissues to remove adhering particles and biofilms, cut with clean scissors into small pieces, placed in solvent-rinsed 500-mL amber glass bottles, and spiked with recovery surrogates, dibromooctafluorobiphenyl (DBOFB) and PCB208. PE were then extracted three times by sonicating in 300 mL of 11939

dx.doi.org/10.1021/es302139y | Environ. Sci. Technol. 2012, 46, 11937−11947

Environmental Science & Technology

Article

Table 1. SPME-Derived CW in pg/L for Samplers Deployed at Three Depths (Mean of Two Fibers) station compounda DDNU

o,p′-DDE

DDMU

p,p′-DDE

o,p′-DDD

o,p′-DDT

p,p′-DDD

p,p′-DDT

depth

BA1C

BA3C

BA4C

BA5C

BA7C

BA7DC

BA8C

BA8DC

BA9C

T11

5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom 5 m below surface mid-depth 5 m off bottom

≤12 ≤14 ≤17 7 21 37 15 27 45 54 170 330